DuDoTrans

• This code is an official implementation of "DuDoTrans: Dual-Domain Transformer for Sparse-View CT Reconstruction" based on the open source ct reconstruction toolbox odl.

Requirements

• platform: linux-64
• python=3.6.13
• CUDA 11.0 or higher

Data Acquisition

• The “NIH-AAPM-Mayo Clinic Low Dose CT Grand Challenge” dataset (NIH-AAPM 2017) could be acquired from here. Note that the download link is updated 2021, and our experimental data is chosen before and differs from the current one, while this doesn't affect the model comparison.
• The COVID-19 dataset is in-house dataset.

Data Preparation (NIH-AAPM)

After downloading the dataset from here, put the train/test data (original dcm files) in the corresponding workdir "path to train" and "path to test".

Training

Run the script on NIH-AAPM dataset.

python main.py

Testing

If you want to test the model which has been trained on the NIH-AAPM dataset, turn off and "trainer.train()" and turn on "trainer.inference()", then download the pretrained model here, put it under the subdirectory ./results/models/, and run the script as follows.

python main.py

Citation

If you use our code or models in your work or find it is helpful, please cite the corresponding paper:

@inproceedings{wang2022dudotrans,
  title={DuDoTrans: Dual-Domain Transformer for Sparse-View CT Reconstruction},  
  author={Wang, Ce and Shang, Kun and Zhang, Haimiao and Li, Qian and Zhou, S Kevin},
  booktitle={International Workshop on Machine Learning for Medical Image Reconstruction},
  year={2022}
}